WEC CONTROLLER, METHOD AND SYSTEM

Abstract

WEC controller, method and system, comprising: a controller (10) for a gyroscopic structure (2) associated with a floating float (3) and comprising an electric converter (9) suitable for converting a rotational energy of said floating float (3) into electric energy, the controller (10) receives as input a perturbed output state (x) comprising an operating variable of said gyroscopic structure (2), determines a drive signal (u) for said electric converter (9), and determines a first signal portion (v) determined using a predictive control model of said gyroscopic structure (2) calculated based on said perturbed output state (x) and a second signal portion (v) determined using a tube convergence calculated on a parametric deviation (r) of said operating variable of said perturbed output state (x). * ), and the parametric deviation (r) is determined by the unperturbed output nominal state (z NP ) of the second signal part (v * ) and

Description

[Technical field]

[0001] The present invention relates to a controller for a Wave Energy Converter (WEC) energy conversion system that includes an electric generator or WEC capable of generating electric energy from sea waves.

[0002] The invention also relates to a method for controlling an electric converter of a gyroscopic structure associated with a floating hull, and to an associated WEC system. [Background technology]

[0003] As is known, wave power is one of the main sources of renewable energy, and in recent years many large plants have been developed and built for the conversion of wave power into electrical energy. Some plants, such as inertial WEC or ISWEC plants, use reactors or PTOs that exploit the inertia of large masses to generate a reaction and extract the power.

[0004] Inertial conversion systems are known that include floating floats fixed to the seabed and equipped with directional gyro converters, each of which is connected to an electrical power generator that is capable of converting the rotational energy resulting from the vibrations of the float and induced by wave forces into electrical energy by the movement of a flywheel.

[0005] In this case, the gyroscopic structure includes a gyroscope, a flywheel associated with the floating body by a suspension device, and an electric converter geared to a rotation axis substantially perpendicular to the main inertial axis of the flywheel. The converter includes an electric motor controlled by a driver / inverter coupled to the rotation axis by appropriate joints and gears. Thus, power can be generated through the electric motor than can be supplied to the electric motor by applying a reactive resistive driving torque acting primarily as a damper, operating in the even quadrant of the VI diagram of the driver / inverter.

[0006] To maximize the extracted power and increase the efficiency of the structure, the opposing resistive driving torque should be appropriately modulated.

[0007] Classical control systems such as PID (Proportional, Integrative and Derivative) controllers are known and widely used in the industrial sector, typically operating as controllers for SISO (Single-Input Single-Output) systems. Although satisfactory under various conditions, systems using PID controllers have drawbacks. Indeed, waves are described in the literature as a random process with statistical distribution characteristics, known as the "JONSWAP distribution", whose parameters depend on the sea area of ​​interest. In the case of PID controllers, the control parameters are updated by a pre-set gain scheduling according to the sea state forecast. Thus, the table values ​​used as control parameters may differ from the values ​​required by the actual wave motion, and a resulting loss of extracted energy is inherent.

[0008] The use of dynamical systems employing controllers using state evolution models is known. Such dynamical systems employ controllers MPC (Model Predictive Control). MPC controllers are described in the article "A comparison of WEC control strategies" by D. Wilson et al., Sandia National Labs, Albuquerque, New Mexico, Tech. Rep. SAND2016-4293, April 2016.

[0009] In its most general form, an MPC controller is based on feedback from the states and a control law that is dynamically calculated by minimizing a suitable cost function to optimize the system states.

[0010] MPC controllers differ from PID controllers in several essential aspects. a) The control laws or functions are based on the solution of the Euler-Lagrange equations which, in classical optimization theory, produce as solution a function of time which occurs as the stationary part (as the superior extremity) of the cost function. b) The cost function contains various terms relating to the states, the input signals, and the kinetic energies associated with the states and the control signals, and is usually a convex function. c) The cost function typically contains terms that are null when the desired state is achieved and terms that are null when the energy of the control action is minimized.

[0011] Additionally, the control laws of the MPC controller may specify dynamic constraints that are used to minimize the cost function. The control laws of the MPC controller are based on the solution of the Euler-Lagrange equations and are essentially a function of time that simultaneously minimizes the error in the required state and the error in the energy committed to obtain this result.

[0012] The use of MPC controllers to control WEC systems is known, by using a state evolution model of the gyroscopic structure. This makes it possible to have a single set of control parameters for all sea wave conditions relevant to the location where the plant is installed. The accuracy of the state evolution model and the associated parameter set estimation affect the performance of the control system.

[0013] In other words, these MPC controllers are optimized for systems with fixed, predefined parameters and are therefore less efficient for systems that are affected by variations in these parameters or by the presence of random disturbances such as sea wave conditions. Similarly, sea conditions that are not taken into account during the calibration of the MPC controller parameters may lead to non-optimal energy extraction conditions, i.e. poor performance.

[0014] A known solution is described in the paper BRACOO G et al.: "Optimizing energy production of an Inertial Sea Wave Energy Converter via Model Predictive Control", Control Engineering Practice, Pergamon Press, Oxford, GB-vol. 96, 17 January 2020 - XP086048062.

[0015] The technical problem underlying the present application is to devise a control of a gyroscope structure, having functional and structural characteristics, that allows the reduction of errors due to the modelling of the waves and of the gyroscope structure, making it possible to maximise the energy extracted, thus overcoming the drawbacks mentioned with reference to the known art. Summary of the Invention

[0016] The solution underlying the present invention is to drive in a constrained manner the future evolution of the operational variables defining the state of the gyroscope structure, improving the robustness and efficiency of the control.

[0017] Based on this solution, the technical problem is solved by a controller defined by claim 1 and by specific embodiments described by claims 2-5.

[0018] The subject of the invention is also a control method as defined by claim 6, and specific embodiments as described by claims 7-10, and a WEC system as defined by claim 11. [Brief description of the drawings]

[0019] Further features and advantages of the invention will become apparent from the following description of a preferred embodiment of the system and its variants, given by way of example with reference to the accompanying drawings, in which: [Figure 1] FIG. 1 shows a schematic of an inertial WEC system (ISWEC) and gyroscope structure. [Diagram 2] FIG. 2 shows a schematic of an inertial WEC system (ISWEC) and gyroscope structure. [Diagram 3] FIG. 3 is a block diagram of a constructed controller in accordance with the present invention. [Figure 4] FIG. 4 shows, in an orthogonal diagram, a schematic representation of the evolution of ideal and actual output states of a tube convergence model applied to a system with two state variables. [Diagram 5] FIG. 5 shows a block diagram of a second embodiment of a constructed controller according to the present invention. [Figure 6] FIG. 6 shows a schematic of a simulation of the time trend of the extracted power of an inertial WEC conversion system with an implemented controller according to the present invention. [Figure 7] FIG. 7 shows some details of the graph of FIG. [Figure 8] FIG. 8 shows some details of the graph of FIG. [Figure 9] FIG. 9 shows some details of the graph of FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] With reference to these figures, 1 shows a schematic of an inertial WEC system, i.e. an ISWEC, comprising a floating float body 3 and a pair of identical and independent gyroscopic structures 2 symmetrically arranged to balance forces with respect to the floating float body 3. Each gyroscopic structure 2 comprises an electric converter 9 suitable for converting the rotational energy of the floating float body 3 into electric energy.

[0021] In the schematic form shown in Figure 1, the floating floating body 3 is substantially symmetrical with respect to a roll axis X and comprises a first gyroscope structure 2, of which a gyroscope 6 is highlighted, and a second gyroscope structure 2'. The second gyroscope structure 2' is shown with a cover protection 7 suitable for covering the underlying flywheel. In the following description, reference will be made to the first gyroscope structure 2 including the gyroscope 6.

[0022] The floating float 3 has a pitch angle δ and JPEG2024527267000002.jpg939 and is formed to enable rotation according to pitch axis Y. In Figures 1 and 2 the wave motion is indicated by an arrow in direction A.

[0023] Although not shown in the figure, in a conventional manner, the floating float 3 is fixed to the seabed and has a roll axis X that is substantially parallel to the wave direction A. The roll axis X is perpendicular to the pitch axis Y. The floating float 3 also has a yaw axis Z that is substantially perpendicular to the plane P defined by the roll axis X and the pitch axis Y.

[0024] JPEG2024527267000003.jpg47166

[0025] The electric converter 9, shown diagrammatically in the figure, includes an electric motor associated with a driver / inverter and is driven by a controller 10 via a drive signal u that counteracts precessional torques to maximise the extracted power.

[0026] FIG. 3 illustrates, via a block diagram, a portion of the inertial transformation system WEC1 of FIG. 1 in conjunction with a gyroscope structure 2 and a controller 10 designed in accordance with the present invention.

[0027] The gyroscope structure 2 is represented by a real plant block 20, i.e. a nonlinear system, including a structure block 21 suitable for representing the real state z of the gyroscope structure 2. The structure block 21 receives a drive signal u and generates a vector containing the unperturbed output state z of the gyroscope structure 2.

[0028] The perturbations and / or disturbances (w) are added to the unperturbed output state (z) to define a perturbed output state x that includes the motion variables of the gyroscope structure 2. The perturbations w include a set of external and internal perturbations to the gyroscope structure 2, such as wave external driving forces that may be filtered by a transfer function and applied to mooring effects and other perturbing elements / forces.

[0029] A controller 10 receives as input the perturbed output states x in order to generate a drive signal u suitable for driving or activating the electrical converter 9. In its most general form, the controller 10 is a TRMPC, an acronym for Tube-Based Robust Model Predictive Control.

[0030] In a first embodiment, the operational variables that define the perturbed output state x may be expressed in vector-matrix form by:

number

[0031] The controller 10 receives a first signal portion v and a second signal portion v * The drive signal u is determined by adding

[0032] The first signal portion v is determined by a predictive control block 13 that includes a predictive control model of the gyroscope structure 2 with the perturbed output state x received as input. In one embodiment, the predictive control model has a cost function, for example, as employed by a conventional MPC controller. In this case, the predictive control block 13 uses a control law whose solution may be based on the solution of the Euler-Lagrange equations by providing a time series that minimizes the error with respect to a required state and minimizes the energy used to realize that state.

[0033] The second signal part v * is determined by a nominal convergence module 18 having a nominal tube convergence. Specifically, the nominal convergence module 18 together with the block 13 enables the convergence of the evolution of the perturbed output states x to the nominal evolution z of the output states according to the TRMPC control.

[0034] The nominal convergence module 18 includes a prescribed gain matrix K by considering the uncertainties of the system and according to a tube convergence prescription for the convergence of the parametric deviation r of the gyroscope structure 2 to a predefined value, such as zero. The parametric deviation r is obtained as the difference of the operating variables of the perturbed output state x with respect to the unperturbed output state z of the gyroscope structure 2.

[0035] Then, the second signal portion v * is obtained as the product of the gain matrix K and the parametric deviation or error r.

[0036] FIG. 4 shows a schematic diagram of tube convergence following TRMPC control of a linear system with two state variables x1 and x2. In the plane of variables x1 and x2, the solid line indicates the nominal state z0-z0 converging to zero. Nspecifies the required trajectory in a time interval T that includes N subsequent steps of length Δt of the evolution of N The triangular shape represents the actual orbit generated by the expansion of the tube. In fact, sections of the tube are highlighted and plotted for each nominal state z0-z N Correspondingly, the allowable actual perturbation state X0-X N represents the appropriate boundary space X' for identifying the N We take into account the time evolution in N subsequent steps up to (t0+T). Naturally, the boundary space X', which in this case is a triangle, is divided into two parts, each of which is a nominal state z0-z N It is assumed that σ includes all possible perturbations w as well as all errors due to uncertainties associated with the model parameters.

[0037] The boundary space X' is the nominal state z0-z N , which is located at the center of the bounded space X'. Of course, the bounded space X' can have a different shape than the triangular perimeter, and the size depends on the strength required for tube convergence.

[0038] The system of Fig. 4 with two state variables, x1 and x2, has a two-dimensional representation in space. Obviously, in the case of the gyroscope structure 2, in the first embodiment, there are four state variables, as shown in the vector matrix 1, and therefore the boundary space X' is four-dimensional. The time interval T is predefined in the design phase, taking into account the dynamic properties of the system to be analyzed, such as the dynamic properties of the waves with reference to the wave period. Of course, the properties of the equipment and / or hardware used are also relevant.

[0039] Nominal state z0-z N are ideal unperturbed conditions and are determined based on an unperturbed model of the gyroscope structure 2.

[0040] According to one embodiment, the gain matrix K is determined by using a procedure following the theory of Linear Matrix Inequalities or LMIs, as described in the paper entitled "Tube-Based Robust Model Predictive Control for Spacecraft Proximity Operations in the Presence of Persistent Disturbance" by M. Mammarella, Capello, Park, Guglieri, Romano, 2018, published on June 1, 2018 by Aerospace Science and Technology - Volume 77, pages 585-594.

[0041] The definition of this region (the "tube") and its width are therefore taken as uncertainty in both the model used in the control and the amount of any disturbances that are not modeled (e.g. mooring effects).

[0042] According to the embodiment shown in Fig. 3, the prediction block 13 is implemented by a nominal unit 14 and a prediction unit 15 arranged in cascade. The nominal unit 14 contains a nominal unperturbed model of the gyroscope structure 2. The nominal unit 14 receives as input the perturbed output state x and also receives as feedback a first portion of the signal v generated by the prediction unit 15, and calculates the unperturbed output nominal state z of the gyroscope structure 2. NP Get the.

[0043] The prediction unit 15 calculates a cost function J that includes quadratic terms for non-quadratic terms related to the state of the gyroscope structure 2, the drive motion, and the instantaneous forces absorbed by the gyroscope structure 2. T The present invention includes a predictive dynamic control model having the following structure:

[0044] JPEG2024527267000005.jpg32166

[0045] Further, according to this embodiment, the nominal convergence module 18 converges the unperturbed nominal state z generated by the nominal unit 14. NP It takes as input and calculates the parametric deviation r. * The second part of is obtained by multiplying the parametric deviation r with the gain matrix K.

[0046] signal v * The second part of the equation is the actual state x0-x of the gyroscope structure 2. N This allows for the modification of the drive signal u while keeping the actual trajectory of the evolution of the gyroscope system 2 more accurately within the bounded space X' or optimal state. This allows the gyroscope system 2 to be optimally controlled even in the presence of external random perturbations w generated by waves.

[0047] A second embodiment is shown in Fig. 5, where the controller 10 determines the drive signal u as an extended drive signal. In the following, only the differences with respect to the previous solution will be specifically described.

[0048] The extended drive signal u is a signal obtained by dividing the first part of the extended signal v by the signal v * and a second portion of

[0049] The prediction unit 15 predicts the extended nominal state z a takes as input and stores the unperturbed nominal state z NP , and the operating variables of the perturbed output state x and the unperturbed nominal state z of the gyroscope structure 2 NP The parametric deviation r is calculated as the difference between the operating variable

[0050] Expanded state z a is a vector that enables block 15 to model and predict the trends of disturbances w while enhancing the prediction of the gyro unit states.

[0051] In this way, the cost function J T The excitation signal generated by the prediction unit 15 by minimizing JPEG2024527267000006.jpg864 allows the determination of an extended drive signal u which is more accurate and therefore makes the control of the electrical converter 9 more efficient.

[0052] The obtained controller is therefore highly robust with respect to the evolution control of the perturbed states. Such control is obtained by both the dual feedback of the perturbed output state x and the dynamic tube convergence model.

[0053] Furthermore, the applicant has been able to verify that the controller 10 so designed enables the gyroscope system 2 to be maintained or returned to a required state even in the presence of uncertainty in the predictive control model parameters of the prediction unit 15 and the nominal unit 14.

[0054] FIG. 6 shows a schematic of a simulation of the trend of extracted power as a function of time for an inertial WEC conversion system using a controller 10 implemented in accordance with the present invention. Power is negative, power extracted. FIGS. 7-9 show some details related to the graph of FIG. 6. The maximum error has been found to be approximately 2% with the following parameter variations: - Gyroscope mass variation (VMG curve): ±10% - Floating mass variation (VMS curve): ±15% - Gyroscope inertial variation (VIG curve): ±15% -Flywheel inertia variation (VIV curve): ±2%

[0055] The charts in Figures 7-9, as specified by Nom, show the trends in extracted power considering the nominal parameters of the model subjected to wave forcing.

[0056] FIG. 7 shows two VMG curves and two VMS curves in addition to the nominal curve Nom. FIG. 8 shows two VIV curves and two VIG curves in addition to the nominal curve Nom. FIG. 9 shows the lower limit L of the extracted power in addition to the nominal curve Nom.I and upper limit L S The curve representing is accentuated by jointly varying the mass and inertia of the floating float 3, the gyroscope and the flywheel.

[0057] The present invention also refers to a method for controlling the gyroscope structure 2 of the WEC system 1 described above, the details of which and cooperating parts having the same structure and function as described above will be indicated with the same numbers and reference characters.

[0058] In particular, the gyroscope structure 2 is associated with a floating float 3 and includes an electric converter 9 for converting the rotational energy of the floating float 3 into electric energy. In its most general form, the method includes Tube-Based Robust Model Predictive Control (TRMPC) of the electric converter 9.

[0059] A controller 10 receives a perturbed output state x which includes the operating variables of the gyroscope structure 2 .

[0060] According to the present invention, a method includes: * The electric converter 9 is driven by a drive signal u obtained by adding a first part of the electric converter 9 to a second part of the electric converter 9 .

[0061] The method devise to determine a first part of the signal v by using a predictive control model of the gyroscope structure 2 calculated at the perturbed output states x.

[0062] Further, the method includes using a nominal convergence module 18 with dynamic tube convergence to determine the second signal portion v * The dynamic tube convergence is calculated by the parametric deviations r of the operating variables of the perturbed output state x. These parametric deviations r are calculated by the relationship between the perturbed output state x and the unperturbed output nominal state z of the gyroscope structure 2. NP The unperturbed output nominal state z NPis obtained by an unperturbed nominal model of the gyroscope structure 2.

[0063] In a first embodiment, a method is devised to determine a first signal portion v using a predictive control model implemented by a conventional MPC controller.

[0064] Furthermore, the method provides for the use of a gain matrix K defined as the "dynamic tube convergence", i.e. the convergence of the parametric deviation r of the gyroscope structure 2 to a predefined value, preferably zero. In general, the dynamic tube convergence deals with linear systems and is shown in a more general implementation in FIG. 4 for a system with two state variables, x1 and x2. Thus, in state space, the gain matrix K is the convergence of the required unperturbed output state z0-z of the gyroscope structure 2. n Perturbed output state X0-X n This enables the deployment of

[0065] The required unperturbed state z0-z n is determined a priori based on a nominal model of the gyroscope structure 2, i.e. considering an undisturbed and linear system.

[0066] In the design phase, the method determines the time interval T, the number of subsequent steps N, as well as each required undisturbed state z0-z n The size and shape of the boundary space X' are devised.

[0067] In this way, in state space, for each perturbed output state x, the parametric deviation r multiplied by the parameters of the gain matrix K maintains the actual trajectory of the gyroscope structure 2 within the bounded space X′ defined for each required state, and the unperturbed required state z0-z n The expansion of n Naturally, for each required unperturbed state z0-z n It is assumed that the bounded space X' surrounding {right arrow over (w)} contains all possible sources of disturbance or perturbation w.

[0068] According to one embodiment, the tube gain matrix K is determined offline by using linear matrix inequality theory. In one embodiment, the matrices A and B used in the classical representation of the linear system in the state space discretization are determined by the cost function J, as detailed in the next chapter. T This is taken into account together with the weighting matrices Q, R and P used in

[0069] In this way, the second part v of the drive signal u * is determined to keep the evolution of the perturbed state inside the tube defined by the boundary space X', with respect to convergence to the unperturbed state.

[0070] According to one embodiment shown in Fig. 3, the method determines the first signal part v via an implementation of a connected prediction block 13 cascading a nominal unit 14 and a prediction unit 15. In the nominal unit 14, an unperturbed nominal model of the gyroscope structure 2 is calculated to obtain an unperturbed nominal state z by using the perturbed output state x and the feedback of the first signal part v. NP is used to generate

[0071] In the prediction unit 15, a predictive dynamic control model of the gyroscope structure 2 is generated based on the unperturbed nominal state z NP and generating the first portion of signal v.

[0072] The predictive dynamic control model is a cost function J that includes quadratic terms related to the state and drive motion of the gyroscope structure 2. T and a non-quadratic term J relating to the instantaneous force absorbed by the gyroscope structure 2. T Includes.

[0073] The method includes: JPEG2024527267000007.jpg864 is the cost function J TThe computation is devised to be used according to an optimization problem determined to minimize

[0074] JPEG2024527267000008.jpg21166

[0075] Further, the method further comprises the step of: NP Using the above, we calculate the parametric deviation r and the signal v * The second part of the above is devised.

[0076] In one alternative, shown in a more general aspect in FIG. 5, the method comprises: a As an input to the prediction unit 15, we devise a method to generate the excitation signal u as an extended excitation signal. a is the unperturbed nominal state z NP , and the operating variables of the output state x and the unperturbed output nominal state z of the gyroscope structure 2 NP The parametric deviation r is calculated as the difference between the operating variable

[0077] Expanded state z a is a vector that enables block 15 to model and predict the trends of disturbances w while enhancing the prediction of the state of the gyro unit.

[0078] In this way, the obtained drive signal JPEG2024527267000009.jpg864 is more accurate and the control of the electrical converter 9 is more efficient.

[0079] The so designed method has achieved the preset goals and objectives by enabling the generation of a so-called robust drive signal u with respect to internal and external perturbations of the gyroscope structure of the WEC system.

[0080] Furthermore, the corrections produced by the controller and obtained by tube convergence implemented in accordance with the present invention allow the gyroscope structure to be returned to an operating state close to the required state even in the presence of uncertainties on the parameters of the nominal units and the nominal models of the predictor blocks, or in the presence of disturbances not previously taken into account and / or modeled.

[0081] (Cost function) Considering the simplified model state shown above, the cost function J of the predictive dynamic control model and associated with the prediction unit 15 is T is detailed according to Eq.

number

[0082] Cost function J T contains quadratic terms related to the state and control or drive actions of the gyroscope structure 2, as well as non-quadratic terms related to the instantaneous absorbed forces. Due to the instantaneous force terms being mixed terms by definition, the cost function J T is not convex and its minimization is sought by determining the drive signal v.

[0083] The power extracted at the kth step and the energy of the state at the kth step zK and the control variable v Ek The energies of are all summed together and calculated over a time interval T that contains N steps.

[0084] The contribution of each term in the calculation of the total cost function is adjusted by the matrix Q associated with the states and the matrix R associated with the control variables. As a weighting coefficient increases, the energy of the associated term is reduced.

[0085] According to one embodiment, the matrix P together with the gain matrix K are calculated according to the theory of Linear Matrix Inequalities or LMIs.

Claims

1. A controller (10) of a gyroscope structure (2) associated with a floating body (3) and comprising an electric converter (9) suitable for converting the rotational energy of the floating body (3) into electrical energy, the controller (10) receiving, at an input, a perturbation output state (x) including operating variables of the gyroscope structure (2), and determining, in the controller (10), a drive signal (u) for driving the electric converter (9), a first signal portion (v) determined using a predictive control model of the gyroscope structure (2) calculated based on the perturbation output state (x), and A second signal portion (v ＊ ), which is determined using tube convergence calculated on a parametric deviation (r) of an operating variable of the gyroscope structure (2), wherein the parametric deviation (r) is calculated as a nominal difference of the operating variable of the perturbed output state (x) with respect to the non-perturbed output nominal state (z ＮＰ ), the second signal portion (v ＊ ) and including, characterized by said determining, controller (10).

2. Considering the time evolution of the required non-perturbed state (z ０ -z Ｎ ) of the gyroscope structure (2), a nominal convergence module (18) comprising a gain matrix (K) suitable for defining the convergence of the parametric deviation (r) to a predefined value, wherein the gain matrix (K) is defined considering the boundary space (X') of each required non-perturbed state (z ０ -z Ｎ ), the controller according to claim 1, characterized in that it comprises said nominal convergence module (18).

3. A prediction block (13) comprising a nominal unit (14) and a prediction unit (15) arranged in cascade, the nominal unit (14) including an unperturbed nominal model of the gyroscope structure (2), The nominal unit (14) receives the perturbed output state (x) as an input and receives the first signal portion (v) generated by the prediction unit (15) as feedback, and generates the unperturbed output nominal state (z ＮＰ ) The prediction unit (15) includes a predictive dynamic control model that receives the non-perturbed output nominal state (z ＮＰ ) as an input, or the prediction unit (15) includes the prediction block (13) that receives the non-perturbed output nominal state (z ＮＰ ) and further the parametric deviation (r) as the input to generate the first signal portion (v). The controller according to claim 1, characterized in that.

4. The nominal convergence module (18) receives, as an input, the non-disturbed output nominal state (z ＮＰ ) generated by the nominal unit (14), the controller according to claim 3, characterized in that.

5. The prediction dynamic control model of the prediction unit (15) has quadratic terms related to the state of the gyroscope structure (2) and the drive signal, and further includes non-quadratic terms related to the instantaneous force absorbed by the gyroscope structure (2), and a cost function (J Ｔ ) including... The calculation for minimizing the cost function (J Ｔ ) is the drive signal determining, wherein the first signal portion (v) is the drive signal by at least one element (v ｉ i = 0..T), said determining characterized by the controller according to claim 1.

6. A method of controlling an electric converter (9) of a gyroscope structure (2) associated with a floating body (3), the electric converter (9) being configured to convert the rotational energy of the floating body (3) into electrical energy, the method providing for receiving a perturbation output state (x) including operating variables of the gyroscope structure (2), and in the method, - driving the electric converter (9) by a drive signal (u) including a first signal portion (v) and a second signal portion (v ＊ ), and - determining the first signal portion (v) using a predictive control model of the gyroscope structure (2) calculated based on the perturbation output state (x), and - the second signal portion (v ＊ ) is determined using the tube convergence calculated on the parametric deviation (r) of the operating variable of the perturbed output state (x), wherein the parametric deviation (r) is calculated based on the operation of the non-perturbed output nominal state (z ＮＰ ) of the gyroscope structure (2), and said determining characterized by a method.

7. - To converge to a predefined final required state (z ｎ ), by means of the nominal model of the gyroscope structure (2), to define the development of the required non-disturbed state (z ０ - z ｎ ) of the gyroscope structure (2), the development of the state being determined over a time interval (T) having N subsequent steps, said defining and - For each perturbation output state (x ０ - x Ｎ ), defining a parametric deviation (r) compared with the corresponding required unperturbed state (z ０ - z ｎ ), and - the actual trajectory defined by the expansion of the perturbation output state (x ０ - x ｎ ) is maintained within a boundary space (X') by multiplying the parametric deviation (r) by the parameters of a gain matrix (K), wherein the boundary space is predefined by the surroundings of the required non-perturbed state (z ０ - z ｎ ), and the multiplying characterized by the method according to claim 6.

8. - starting from the perturbation output state (x) of the gyroscope structure (2) and the feedback of the first signal portion (v), the unperturbed nominal state (z ＮＰ ) is generated by using an unperturbed nominal model of the gyroscope structure (2); - To generate the first signal portion (v), using the predictive dynamic control model of the gyroscope structure (2) based on the received non-disturbed nominal state (z ＮＰ ), or, to generate the first signal portion (v), using the predictive dynamic control model of the gyroscope structure (2) based on the received non-disturbed nominal state (z ＮＰ ) and further based on the parametric deviation (r). characterized by the method according to claim 6.

9. The predictive dynamic control model devises a cost function (J Ｔ ) that has a quadratic term related to the state of the gyroscope structure (2) and the drive signal (u), and a non-quadratic term related to the instantaneous force absorbed by the gyroscope structure (2). - drive signal The cost function (J Ｔ ) is used for the calculation according to the optimization problem of minimizing - the drive signal at least one of the elements (v ｉ i = 0 ,..., T) to determine the first signal portion (v), and characterized by the method according to claim 6.

10. - receive the perturbation output state (x) as an input and receive the first signal portion (v) as feedback, thereby providing a nominal unit (14) suitable for defining an ideal unperturbed output state (z ＮＰ ) - calculating the parametric deviation (r) by using the ideal non-disturbed output state (z ＮＰ ) generated by the nominal unit (14); and characterized by the method according to claim 6.

11. - a floating body (3), - at least one gyroscope structure (2) associated with the floating body (3) and comprising an electric converter (9) suitable for converting the rotational energy of the floating body (3) into electrical energy, - In a controller (10) that receives, as an input, a perturbation output state (x) including an operating variable of the at least one gyroscope structure (2), characterized in that it is configured according to any one of claims 1 to 5, the controller (10) and including, a WEC system.

Images